Help - Search - Members - Calendar
Full Version: Equalizer: Unexpected behaviour
Hydrogenaudio Forums > Hosted Forums > foobar2000 > Support - (fb2k)
tigre
Hi.

I've been curious about how foobar2000's equalizer behaves on 24bit/96kHz files, so I created some equalized files using diskwriter (fb2k 0.7 final). Everything seems fine except for the highest band (labeled as "20 kHz"). No matter what it is set to, all frequencies between ~11.9 kHz and 48 kHz are amplified by the value that has been set for "14 kHz" band.
I've tested this with CoolEditPro's fequency analysis, only with 24/96 files so far.

CU tigre
tigre
I've just tested the EQ again, this time fb2k 0.7.1:

Using CoolEditPro I created 3 .wav files containing white noise at 44.1, 48 and 96 kHz sampling rates (all 16 bit stereo). Then I loaded the .wav files in fb2k; all DSPs disabled besides Equalizer.
EQ settings: 55Hz -> -20dB; 77Hz -> -15dB; 110Hz -> -10dB; 156Hz -> -20dB; 220Hz -> -15dB and so on

Then I used fb2k's diskwriter to write the equalized noise to .wav files (no resampling) and analyzed the files using CEP's frequency analysis.

Result:
The "20 kHz" Band has still no effect. Everything > ~11.9 kHz seems still belong to the "14 kHz" Band as described in my post before. This error occured on all sampling rates tested.
mezenga
I did a much less elaborated test but got similar results...

Equalizer was the only active DSP.
Simple spectrum was used to see resulting output.
All bands were seted to -20dB.
Choosed a music that make regular use of all bandwidth.

Getting individual bands (one by one) to +20dB result in that band being cleary ploted on the spectrum except for 14 and 20 kHz.

The 20 kHz band didnīt change a single pixel on the spectrum while the 14 kHz band shows itself wider than 7 or 10 kHz cleary taking the 20 kHz band.

So, my amateur conclusion is:
The 14 kHz band high pass filter isnīt working and the 20 kHz band isnīt working at all. sad.gif

I hope that Garf or Peter can look this issue.

edit: before you try this, keep in mind that I did a graphical analyzes only. I higly recommend you to disable the sound output in the windows mixer to avoid hurting your speakers...
kode54
I found this bug this morning and it has already been fixed and will appear in the next version.

If you have the tools to compile the component yourself, here is the fix:

Go to line 144 of Equ.cpp, which should read:

CODE
(*pp)->upper = i == NBANDS-1 ? fs : bands[i  ];


Change it to read:

CODE
(*pp)->upper = i == NBANDS ? fs : bands[i  ];



The same bug exists in dsp_supereq 0.03.
mezenga
Silly bug, silly fix ... tongue.gif

Thanks kode. biggrin.gif

And thanks to tigre for noticing this smile.gif
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.